Digital Signal Transmission

[TomJ]

So one question about the reason of the sound improvement.

How could the software matter as long as the data is being properly transmitted from the receiving TCP / IP stream to the USB port? Sound differences can then only be caused by electromagnetic interference. I use a USB isolator that makes a transfer (asynchronous) to my DAC impossible.
Does anyone have an explanation for this?
In my opinion, there is currently so much nonsense in the forums in the field of digital singnal transmission.

[TomJ]

In my opinion, the most credible explanation is most likely to be found in the field of autosuggestion.

[TomJ]

As I see that my questions also opens in this forum cans of worms with never ending discussions please ignore this.

 

[TomJ]

19 minutes ago, Superdad said:

It is always so tempting! 

 

Its so annoying - there are so many myths about digital audio, that you can not distinguish whether it is true or only justification for the price paid.

[TomJ]

I hope we don’t need to discuss jitter over TCP / IP (which all streaming systems except Apple use today).
So the data is always the same, regardless of whether it comes to the endpoint via WiFi, LAN or fiber.
So only interference could change the sound.
I once connected a power bank to the streamer and the unclean electricity from the power bank was initially interpreted positively as more space.
With the isolator between my streamer and the RME ADI 2 DAC, this effect is over. Everything that is done in front of the isolator no longer has any (audible) influence on the sound.
I'm speculating now that the alleged sound improvements are actually a deterioration due to HF interference, which, if you try out intensively, have the right dose and please. This means that local files may sound better to some, as interference from hard drives, ssd readers, etc. are interspersed, but this does not correspond to the original sound.

[TomJ]

Just to get me right: I am happy with the isolator and am not looking for any "improvements" through bad electricity, like the example with the power bank. Of course, interference can still jump from one side of the isolator to the other, but the influence of HF and noise is minimized. This means that changes in the digital chain are no longer audible. So all so-called sound optimizations in the digital chain without isolator are just a gambling for the right dose of HF and noise, but then have nothing to do with the sound of the original recording. The "more space" or darker background e.g. through a special LAN cable is not available on the source, but is added incorrectly. To save myself such gambling, I am very happy with the isolator.

[TomJ]

To me whats been recorded is in the data. And to transform the data back to music its the job of the dac. If you are not happy with the sound of the dac, then I would buy an other dac rather than fiddling with cables, power supply of routers and so on.

So if the data is sent correctly to the dac (this is what RME dac can verify with a build in bit perfect test), I think the dac can do the best job by eliminating noise and HF. If you will add some flavor to the sound whats not on the recording with cables and so on, the isolator is not for you.

[TomJ]

7 hours ago, fas42 said:

 

Every half keen audiophile does that already - except, it's called tweaking .. remember, we're talking external stuff. Hmmm, vibration might be a problem, okay, we'll explore mounting it on different platforms; power supplies aren't good enough, okay, we'll add external mains power conditioning to compensate for the lack of internal cleaning up of such; it's sensitive to the nature and quality of input waveforms, okay, we'll condition those in myriads of ways, to address the fact that the internal engineering is not robust enough to cope with whatever comes its way ... and on it goes ...

 

If you're an audio enthusiast, you want better sound ... normally 😜. Therefore, a  good use of your time is to try things to "make it better" ...

I agree with you regarding the analogue part of the chain. But regarding the digital side the goal must be to transport the data without any influence to the DAC. And cable-sound or sound of a router should not be the target. But I can still here slightly differences of USB cable from Isolator to the DAC. This could be because different cable handle HF and noise different. Best way would be, that the Isolator would be integrated on the board of the DAC.

[TomJ]

34 minutes ago, fas42 said:

 

Yes, the isolation mechanism should be part of the DAC - the concept is that the converter is able to handle any digital input which is measurably correct, even though the signal lines, waveforms are as noisy as hell, with zero audible impact. That is just good engineering, and should be the end goal.

To transport the true data to the DAC is an IT task. USB asynchronous can do best with lowest jitter rate (if any). So I do not understand, why many so called Audiophiles prefer an unstable SPDIF and setup masterclocks for thousands of euro with the end, that transport of data is only near the quality of USB. I think many people try to use their knowledge from the analogue domain to optimize the digital side - but these are completely different technologies.

[TomJ]

37 minutes ago, sandyk said:

 That statement shows a complete lack of knowledge and Anti  Subjective Bias .

 

 See also actual USB measurements :

https://www.computeraudiophile.com/forums/topic/24275-ac-filtering-grounding-boxes-linear-psu-and-balanced-power/?page=68 posts by YashN

I was talking about USB as a data transport, not as a power source for a dac, what was the focus of the measurements of your link.

Maybe it is known, that there are DAC not using the power of USB.

Lack of knowledge?

This is what I mean that there is so much confusion around digital audio.

I talked about the data precision of USB in case of feeding a LPS powered DAC  (my RME) and you show me the dirt of USB power and all the people think USB is eval.

[TomJ]

3 hours ago, PeterSt said:

 

Tom, USB *is* evil. Trust me. It can be turned into an interface with lower jitter than SPDIF, but this is so difficult that it took 6 or 7 years after asynchronous USB was "invented" for audio to get there by one or two DAC manufacturers.

 

And a hint for you (it could silence you somewhat 🤫): USB implies a relatively very high 8KHz signal, which is right in the audio band. It is pure distortion. Here too, it is the most hard to tame, but it can be done to at least something which should be inaudible, but which will still always be measurable.

In itself this signal implies (data-) jitter, because it is an oscillation beating with the DAC's audio clock.

Hi Peter,

I always find it funny how you are immediately dismissed as ignorant in these audiophile forums when you ask questions that do not fit into the worldview.
I am not an electrical engineer and I will not become one.
I have spoken of it myself that I am amazed at the amount of effort that is sometimes put into it and I do not understand it because it does not reveal itself to me with my knowledge.
I keep getting answers that bad jitter is to blame and that jitter is also the cause of the deterioration of the sound over the network - which is absolute nonsense, since all systems use TCP / IP.
For your information: I have a streamer that is operated with an LPS (I am aware that the USB power is otherwise useless). The streamer streams via USB to an Intona isolator, which is passive and has no other power supply, such as the Uptone Regen and therefore does not worsen the power situation.
Between the isolator and the RME ADI 2 DAC is a NEO Oyaide d + Class S cable, on which the + 5V contacts are masked on both sides.

Which DACs are you talking about that have overcome the USB problem? 

[TomJ]

1 hour ago, pkane2001 said:

 

RME ADI 2 DAC is pretty much immune to the usual noise from USB line, so ignore any potential issues with PC case rivets or 8KHz noise :)

 

I have the V1 RME ADI 2 DAC, which looks also quiet perfect with "eval" USB:

 

 

@PeterSt:

Here is a comparison of S/PDIF vs. USB - USB is the clear winner:

https://www.audio “science” review/forum/index.php?threads/battle-of-s-pdif-vs-usb-which-is-better.1943/

 

But now I must read next, that this measurement is garbage because not done with the super duper master clock and done with garbage DACs and not with 10T$ DACs and you must use your ears, bla bla bla. This is what I meant with the myths around digital audio. The measurements shows, that USB is better, but If you post this in forums you are directly portrayed as a fool without a clue. 

 

Here are the Measurements S/PDIF vs USB:

 

 

 

 

[TomJ]

An other thought I want to share:

Maybe USB is not so popular because you can't fumble around here enough and you can't make such a big secret about the great sound optimizations. Actually a big annoyance for all "magicians" of the great sound. 

[TomJ]

@PeterSt

Sorry if I was a little too rude. I have just come back from endless discussion in another forum regarding digital audio and it always ended with the same esoteric statements and "I can hear the difference" without any explanation. I have great respect for your work!

With "passiv" I used the wrong word for the Intona. I know, that it use the USB  power.

Daniel Stämmler has not yet developed a passive isolator.

 

And TCP/IP was not mentioned as a solution for the USB topic. I come from IT and I'm a bit tired of the discussions about jitter effects on sound in network transmission. There is a solution for this and that is the TCP / IP protocol, which compensates for possible errors on lower ISO layers. But if I'm wrong here, I ask for clarification.

 

@Superdad

Same to you - neither I “just fell off the turnip truck” or I am trolling.

With regard to the popularity of Amir, there seems to be a need for transparency and comparability. There are apparently a lot of people who are fed up with just esoteric talk without explanations. If you have better measuring devices and methodologies, then I think measurements with this would be welcome. It is too easy to just discredit Amir.

 

[TomJ]

19 minutes ago, Superdad said:

 

Okay, let's see how few words I can condense this into:

 

Like everything else in this digital world, Ethernet is voltages into chips that run from clocks.  And not even binary at that: 100Mbps is 3 voltage levels, Gigabit is 5 voltage levels.  Just like everything else, the clocks are voltages with a rise time.  And clock voltages are referenced to some 0-volt "ground" plane.  Allow ground plane noise to enter an endpoint device (one that ultimately has a path to an audio rate master clock pin) and you end up with a bouncing "ground." Since the clock is not a perfect zero-rise-time square wave, movement of the 0-volt reference is laterally moving (in time) the clock voltage threshold of whatever chip is running off it. Jitter.

 

Plus every chip--be it a PHY, MAC processor, FPGA, digital isolator, or flip-flop--generates its own noise (often via high-speed current draw spikes) and jitter.

 

Separately, the causes of the above--ground-plane-noise/clock-threshold jitter--range from common-mode noise and leakage currents to upstream phase-noise (jitter) induced perturbations. And all that stuff adds up and becomes audible.

 

Not going to go down the rabbit hole with anyone about why this above eludes standard FFT measurements at the analog output. That's a separate debate for which I will not take the bait.  

 

Funny thing is: DAC and other digital engineers (especially if you get into RF, telecommunications, and microwave engineering) deal with the ground-plane and clock issues in their designs as a matter-of-course. Really nothing new of controversial about any of it.  What the armchair engineers, network guys--and even those designers who think nothing gets into their box--are missing is that immunity from electrical factors does not automatically come with every packet data interface, be it USB or Ethernet.

 

So we're talking about jitter that happens in the DAC when converting from D to A due to noise from the ethernet line? Not about jitter that leads to wrong data? I would be suspicious of this, but it is understood that way by many and therefore I am giving the TCP / IP argument.

 

[TomJ]

12 minutes ago, Superdad said:

Bingo!  What goes on--and what our own highly unique EtherREGEN is all about--has zero to do with changing the bits. Never has.  A $20 switch will transmit the same data. It is the secondary electrical factors at play--which are affecting various stages downstream.

The problem I have, that many people i discuss with how a network device can improve the sound dont understand that even with streaming services over the internet, there is no missing bit when the stream arrives in their streamer. They argue with UDP protocols and missing bits and this should be the reason why better "audiophile" switches improve the sound. And i am sick about these discussions.

 

If I am using a network isolator, then the noise will be lowered -right?

 

I use such an isolator in front of my streamer and an isolator between streamer and dac.

Isn't this enough to keep the noise outside of the dac?

 

 

[TomJ]

Just now, sandyk said:

 The obvious answer then is to ignore this forum and listen to, and view the world according to Amir and his purely measurement brigade in  A.S. where you presumably are already a devoted member ?.

So if there is no consensus in this forum that the data will be sent correctly over the network to the network end device and that this is the basis for discussion here in this forum, then you are right.
I'm not talking about the fact that a switch cannot increase the sound quality, but that it does not change the data less than 20 $ switches.

[TomJ]

14 minutes ago, Superdad said:

Like going to the doctor and saying: "If I am taking medicines then sicknesses will be better, right?"

Or to the chef and stating "My food will be better with spices."

But the problem then seems to be more about making the diagnosis. But how, without measurements? Seems to be a gamble.

[TomJ]

So thank you all so far.

I'll make a topology drawing so that we can work our way along this. (will take a while)

 

Here is my situation again:
I can hear the difference with Intona. Even without the intona, the RME shows "Bit Perfect" - no tdata transport problem. So something positive has to happen with the intona. Since there cannot be any missing or faulty bits, it can be assumed that it is noise that is being isolated. I can also hear slight differences in the sound of the USB cable connected to the RME. The Neo Oyaide was the best so far - slight emphasis on height, but I like that. So the cables handle noise or what changes the sound in different ways.
I can also hear an improvement with the Network Isolator on Ethernet. This is stronger with my speaker setup with KEF LS50 Wireless, which I operate via Ethernet, than with my headphone setup with the RME. So here too the conclusion that it must be noise or something else that the isolator is blocking on ethernet, since it cannot be the manipulation of bits.

Even if this cannot be measured, I can hear a difference - but neither do I want to rule out that this is autosuggestion when investigating the cause.

[TomJ]

To get an better overview my topology. 

Since I mainly stream, here is the path of the data from the streaming provider to the speakers / headphones.
I don't hear any difference between streming from providers or from my local server. 

 

[TomJ]

So the goal is to prevent noise coming into the dac -right?

This should be the task of the isolator. But I have understood, that noise can "jump" from one side to the other in the intona -right? And maybe the USB cable #9.1 catch some noise from the environment.

But if the isolator would eliminate the noise, than I don't have to care about audiophile server or audiophile switches - right?

How about changing the isolator and cable (9.1 + 8.1) with an fiber usb extender like this: https://www.amazon.com/LINDY-42683-Optisch-Elektronik-Schwarz/dp/B00WJ8DNUK/ref=asc_df_B00WJ8DNUK but only shorter.

Then I only have to care about a clean power supply for the extender - right?

 

 

 

 

[TomJ]

58 minutes ago, PeterSt said:

 

If all were that simple ...

First off, all which is glass (fiber) always sounds worse. This will be related (unproven) by the processing which is required to (in-DAC) convert back to copper. Throughout time a few in the group at Phasure (including myself) tried this (IIRC this was always about Adnaco).

OK - So Toshlink always sound worser than USB.

Is this also the case for Ethernet to fiber and back in your opinion?

 

Quote

Tom, let's keep in mind that once you are as far that you tend to work on phenomena like blocking noise, the better approach really would be the ISO-Regen. That isolates and re-generates a clean signal at the same time. Regarding this, try to grasp that the bits can be captured 100% without error by such a device, and including isolation from the dirty (PC) side, close to the DAC the signal can be regenerated as clean as possible, preferably with a clean PSU indeed.

But ... welcome to the world of USB tweaks after all ?

 

So with ISO-Regen there is no need for audiophile switches and servers?

 

[TomJ]

2 hours ago, PeterSt said:

If that would be the last element in the chain, then very far, theoretically, it would be correct. But it is not at all because my software still happily change the sound as always, without and with ISO-Regen. And thus the answer is (sadly) No - that need remains (but mind my (non-)opinion on switches).

If your mentioned "server" is the audio playing PC, then that is the most crucial of it all, these days. However, please notice that this is in the context of not-streaming from the Internet and only local playback (you mentioned this as making no difference for you).

 

As you can see in my topology the Logitech Squeezebox streamer is the audio playing device - its not a pc. No harddisk / ssd, wlan or other stuff like in pc. The streamer always gets his stream from the LMS Server over Ethernet with LAN isolator - from local files stored on the server or from a streaming provider.

The server transcode the stream (local files or streaming provider) into pcm before streaming to the clients (Logitech Squeezebox or KEF LS50). So if i say local playback it means streamed from my own server, not from the streaming provider like Tidal/Qobuz.

 

I am a bit confused (maybe this is strategy?).

If a device like ISO-Regen regenerate a clean USB signal for the DAC, why should one invest in other stuff like audiophile streaming client, ethernet cable, switches, servers with "low noise"? So a ISO-Regen destroy the business of all these companies? Ah, no - everything matters! But why, if the dac gets the cleanest signals he can get?

 

 

[TomJ]

So I sum up:

• Data is not changed on its long way across networks. The correct data always arrives at the network end point (e.g. network streamer)
• Jitter problems in switches do not cause "worse" data, but produce "noise".
• "Noise" can "jump" from one side to the other in a device even though there is electrical isolation.
• Although Ethernet is always galvanically isolated on the data path (if shielding is not used), "noise" is transmitted from the previous one.
• In order to transfer the data from the network endpoint to the DAC, USB is superior to the other transport options.
• Jitter can worsen the sound in the DAC when converting from digital to analog. These can be caused by "noise" entering the DAC via electrical channels or by EMI in the environment.
• Sound improvements or deteriorations through "noise" via the digital transmission path cannot (or only with great difficulty) be measured.
• A USB isolator does not eliminate all of the "noise", but perhaps only 60%.
• The aim must be to keep "noise" out of the DAC.

 

If noise is the cause of all the evil and the effect cannot be measured, why is the cause not simply measured ("Noise")?

Why not stated the noise level for products done by measurements , but always only talked about sound-enhancing "Low Noise" products without giving any values?

[TomJ]

Another Point:

 

• If the network "end-point" have to much work to correct false data (TCP/IP), it produce more noise by calculation. (Have there been any measurement about corrupt packages at network endpoints (e.g. streamer) yet?  This would increase the reputation of so called audiophile switches, but "you can hear it - no need for measurements"😁)

 

So there are things you can measure regarding sound quality:

• Measurement of noise level entering the DAC or of the ethernet path.
• Date integrity of the network transport. (should be easy, because this is daily business of IT guys). If a super-duper clock in a switch will be worth, then an improvement should be measured.

 

Why cant I see discussions based on those facts and only read about magic stuff is going on?